Noah Faro

At @farsight_ai, we got tired of people explaining their workflows to AI - what they should look like, what they should include, how things are connected. Today we’re launching Freeform to fix that: SOTA, simultaneous large-scale generation across PowerPoint, Excel, and Word. Creating 100+ page slide decks, multi-sheet interconnected excel models, and robust and lengthy narrative documents from scratch is a thing of the past now. With Freeform, you can input a single prompt, and our system rationalizes across all of your data to understand not only what information is needed to execute something, but also what set of materials and outputs are required to complete it end-to-end. Freeform is already taking real, meaningful work off our customers’ plates. It understands how teams actually work, why they do things the way they do, and matches their way of doing things exactly. This is the first solution in finance that completely removes the mental burden of using AI, allowing it to do the work completely for you, exactly how you would do it. This fundamentally changes how financial work gets done. Extremely proud of what the team has built here. Comment “Freeform” to get a demo.

@shensi Of course application layer companies will do this, but I wonder if there will be an incentive in the future for Anthropic or OpenAI to do this themselves in CC & Codex by default. Would combat usage drops of higher thinking models due to cost

the cost of fable is going to make smart model routing impossible to ignore

@scaling01 It may cost less than half of GPT-5.5 theoretically but I asked it to do a fairly simple task that analyzes logs for me and suggest improvements in a system, and I went through 100% of my CC usage in < 30 minutes. With GPT-5.5 xHigh, I only used 8% of my usage and it took 8 mins

Fable 5 is also by far the best computer use model according to Stagehand Agent Evals it also costs less than half of GPT-5.5

Feel like inevitably (if they aren’t already) @mercor_ai or @joinHandshake are already helping give them manually labeled design data. @figma though has a really interesting opportunity to hedge AI design development here if they can make something actually work with one of the labs. A bit out there, but would immediately king-make one of the labs in terms of model taste. If they can sell their design data within Figma Community to a lab in exchange for either of: * The lab not being able to release a design IDE * Giving them a revenue share of any design-specific vertical applications they release … that would be a great hedge. Otherwise, the data will get to them eventually, and they will do what they want with it when their models finally align with human taste

i think models will be good at design once frontier labs can trick people like me into sitting down and labeling data for them all day for a year.

temporal is the sauce

@ajambrosino Make a code editor embedded into Codex and watch the world flip upside down

not to worry– we're also taking Codex to the next level for software development. what should we focus on that we aren't already?

Purely a different GTM strategy - OpenAI w/ Consumer v Anthropic w/ Enterprise. It generally explains the difference in their initiatives too: Claude Design, Claude Cowork, Claude O365 Add-Ins vs ChatGPT’s Ads, GPT Images 2.0, and now Sites for E2E website creation. Now OpenAI is making a big enterprise push - and I think it’s easier to break into enterprise vs consumer given both have a 1T val. I think we’ll see this MAUs remain being a very large gap, but I don’t think this is indicative of company performance

Claude apparently has ~5% as many MAUs as ChatGPT. That's wild.

@nosimus1 @scottastevenson René Descartes is that you? Great point actually lol

@noahfaro @scottastevenson Saying "I don't know" also agrees with him, since every conscious entity knows for sure that it is conscious.

AI is obviously not conscious

Aside from model-agnostic horizontal companies, this is also why vertical-specific AI companies are important. Not having to lock your business in to a single AI model provider ensures that you take full advantage of ALL AI advancements and ensures you don't get price gouged

@benhylak announcement format looking like a festival lineup also caters to the mega-consumers that would both use a personal shopping and go to festivals. makes a lot of sense

anyone clowning on this is very dumb. she's building a personal shopping assistant. these are very useful people to have on your cap table.

@scottastevenson jk. this is how many interactions I had to give to Claude to get it to just say no instead of "I don't know" lol. Your debate remains open

@scottastevenson Claude agrees with you

@pranaveight Do you think that Codex usage will go up dramatically due to the merge of UX? How are you all going to push users towards Codex that are currently unaware of it? Feel like people that would make use of Codex that are ChatGPT subscribers already use it, so curious for thoughts

The Plan:

@pranaveight @theo never heard anyone say “S-W-E” at all until now, sweebench is the only correct answer - all my love tho @theo

Wait .. guys.. Do you say S-W-E Bench or sweebench? Everyone at OpenAI is in the latter camp but TIL that @theo is in the former!

@diegocabezas01 don’t think the first part could’ve been more clearly a jest

@noahfaro lol but both ChatGPT and Codex are from OpenAI

Codex will be merged into ChatGPT

@thatguybg Better than #000000

if you use this color for your dark mode I hate you

@TheRohanVarma Been a long time coming tbh given the current coding agent capabilities. Super interested to see how Loveable, Replit, even Vercel react. Congrats Codex team!

We just launched Sites into Codex! Software creation was always about more than writing code. Sites in Codex fundamentally gives the power of end-to-end software creation to every user, no matter their technical fluency. These Sites are fully deployed to a URL, private to workspaces, come with authentication, can have static files, and can store dynamic data in databases. It is in preview for business and enterprise teams and will be rolling out to all workspaces over the next day. Give it a try by typing @ Sites into Codex and ask it to build anything! This project took a massive amount of effort across hundreds of people at OpenAI - proud that we were able to get this out and excited to see what you all build with it!

@thatguybg caffeinate -i

sfo-ing so hard

@thsottiaux OpenAI’s finance team to @thsottiaux

Five million users would agree. Resetting the limits tomorrow morning to celebrate. Time to go /fast

@metapreston They haven’t rearchitected the Siri model since it first came out - I think that non-hardware large r&d is just not worth the investment for them

APPLE IS A COMPUTER COMPANY